TranscriptGeneMap transcriptToGeneMapFromFasta( const std::string& transcriptsFile ) {
using std::vector;
NameVector transcriptNames;
NameVector geneNames {"gene"};
vector<bfs::path> paths{transcriptsFile};
StreamingReadParser parser(paths);
parser.start();
ReadProducer<StreamingReadParser> producer(parser);
ReadSeq* s;
// while there are transcripts left to process
while (producer.nextRead(s)) {
// The transcript name
std::string fullHeader(s->name, s->nlen);
std::string header = fullHeader.substr(0, fullHeader.find(' '));
transcriptNames.emplace_back(header);
producer.finishedWithRead(s);
}
// Sort the transcript names
std::sort(transcriptNames.begin(), transcriptNames.end());
// Since we have no real gene groupings, the t2g vector is trivial,
// everything maps to gene 0.
IndexVector t2g(transcriptNames.size(), 0);
return TranscriptGeneMap(transcriptNames, geneNames, t2g);
}
Index findTranscriptID( const std::string &tname ) {
using std::distance;
using std::lower_bound;
auto it = lower_bound( _transcriptNames.begin(), _transcriptNames.end(), tname );
return ( it == _transcriptNames.end() ) ? INVALID : ( distance(_transcriptNames.begin(), it) );
}
